Sean Trott

Assistant Professor in Psychology at Rutgers University.

[James Michaelov @ ACL 🌴 @jamichaelov]

Seems like a good time to share our new preprint about model openness! (with @linguist_cat @tylerachang @PamRiv1 @SamuelTaylorCS @camrobjones @Sean_Trott @roger_p_levy Ben Bergen @drmaltman):

[Sean Trott @Sean_Trott]

@dioscuri Have you read Wizard and the Prophet? Or Small is Beautiful? My sympathies are ultimately more "wizard" than "prophet", but both those books present a philosophy that's arguably antecedent to degrowth (and that I find broadly legible/compelling).

[Sean Trott @Sean_Trott]

@amuuueller @_shruti_joshi_ This looks very cool! Figuring out the functional scope / generalizability of candidate mechanisms seems like one of the main challenges/problems for the field.

[Sean Trott @Sean_Trott]

@raphaelmilliere @KordingLab Also curious!

[Sean Trott @Sean_Trott]

@mariabrbic Very cool!

[Sean Trott @Sean_Trott]

@LedermanHarvey Definitely, I think it would avoid a lot of confusion / misunderstanding!

[Sean Trott @Sean_Trott]

@LedermanHarvey I also think the broader point you’re making about where it makes sense / matters to draw boundaries is a really good one.

[Sean Trott @Sean_Trott]

@LedermanHarvey I agree, but maybe a more reasonable version of this pushback can be grounded in the difficulty in establishing functional equivalence in the first place; eg, I’ve long suspected that what seems like axiomatic skepticism is sometimes skepticism about behavior generalizing.

[Sean Trott @Sean_Trott]

@najoungkim Thanks!

[Sean Trott @Sean_Trott]

Very excited to announce that I'll be starting as an Assistant Professor in the Psychology department at Rutgers University-Newark in January 2026!

[Sean Trott @Sean_Trott]

@tallinzen Thank you!

[Sean Trott @Sean_Trott]

Feel free to reach out here or elsewhere if any of this interests you!

[Sean Trott @Sean_Trott]

I'm also really interested in epistemological questions about how best to study LMs, relating primarily to construct validity (are the tasks we use appropriate?) and external validity (are the findings we obtain generalizable, and to which "populations" of LMs?).

[Sean Trott @Sean_Trott]

(Thanks to @camrobjones, @PamRiv, Oisín Parkinson-Coombs, and Kola Ayonrinde for valuable comments on various iterations of this work!)

[Sean Trott @Sean_Trott]

Another problem is that this is simply very hard to implement: we don't have random seeds for most models! Indeed, the problem is even worse: available models are *not* a representative sample of possible models! But this should just make us more cautious about our conclusions.

[Sean Trott @Sean_Trott]

In general I think there's tons of interesting work to be done exploring what *kinds of claims* generalize across *which kinds of model instances*!

Paper link here: https://t.co/BzVJlmMlgG